/* @func embPatternSeqCompile *************************************************
**
** Adds compiled pattern into AjPPattern.
**
** @param [w] pat [AjPPatternSeq] Pattern for compiling
** @return [AjBool] True, if compilation succeeded
** @@
******************************************************************************/
AjBool embPatternSeqCompile (AjPPatternSeq pat)
{
    AjPPatComp embpat;
    AjBool embType;
    AjPStr pattern = NULL;

    ajStrAssignS(&pattern,ajPatternSeqGetPattern(pat));
    ajStrFmtUpper(&pattern);
    ajDebug("embPatlistSeqCompile: name %S, pattern %S\n",
            ajPatternSeqGetName(pat),pattern);

    embpat = ajPatCompNew();

    if (ajPatternSeqGetProtein(pat))
	embType=ajTrue;
    else
	embType=ajFalse;

    if (!embPatGetTypeII(embpat,pattern,
			 ajPatternSeqGetMismatch(pat),embType))
    {
	ajDebug("embPatlistSeqCompile: Illegal pattern %S: '%S'\n",
		ajPatternSeqGetName(pat),ajPatternSeqGetPattern(pat));
	ajPatCompDel(&embpat);
	ajStrDel(&pattern);

	return ajFalse;
    }

    embPatCompileII(embpat,ajPatternSeqGetMismatch(pat));
    ajPatternSeqSetCompiled(pat,embpat);
    ajStrDel(&pattern);

    return ajTrue;
}
Esempio n. 2
0
ajuint ajPatlistSeqDoc (AjPPatlistSeq plist, AjPStr* Pdoc)
{
    AjPPatternSeq pat = NULL;

    ajFmtPrintS(Pdoc, "%-12s %8s %s\n",
		   "Pattern_name", "Mismatch", "Pattern");
    while (ajPatlistSeqGetNext(plist, &pat))
    {
	ajFmtPrintAppS(Pdoc, "%-12S %8d %S\n",
		       ajPatternSeqGetName(pat),
		       ajPatternSeqGetMismatch(pat),
		       ajPatternSeqGetPattern(pat));
    }
    return ajListGetLength(plist->Patlist);
}
void embPatternSeqSearch (AjPFeattable ftable, const AjPSeq seq,
			  const AjPPatternSeq pat, AjBool reverse)
{
    const void *tidy;
    ajuint hits;
    ajuint i;
    AjPPatComp pattern;
    EmbPMatMatch m = NULL;
    AjPFeature sf  = NULL;
    AjPSeq revseq  = NULL;
    AjPList list   = ajListNew();
    AjPStr seqstr  = ajStrNew();
    AjPStr seqname = ajStrNew();
    AjPStr tmp     = ajStrNew();
    ajint adj;
    ajint begin;
    AjBool isreversed;
    ajint seqlen;

    seqlen = ajSeqGetLen(seq);
    if(!seqlen)
        return;

    isreversed = ajSeqIsReversedTrue(seq);

    if(isreversed)
	seqlen += ajSeqGetOffset(seq);

    begin = ajSeqGetBeginTrue(seq);
    adj = ajSeqGetEndTrue(seq);

    if(!ajStrGetLen(featMotifProt))
        ajStrAssignC(&featMotifProt, "SO:0001067");

    if(!ajStrGetLen(featMotifNuc))
        ajStrAssignC(&featMotifNuc, "SO:0000714");

    ajStrAssignS(&seqname,ajSeqGetNameS(seq));
    pattern = ajPatternSeqGetCompiled(pat);

    if (reverse)
    {
        revseq = ajSeqNewSeq(seq);
        ajStrAssignSubS(&seqstr, ajSeqGetSeqS(revseq),
			begin-1,adj-1);
        ajSeqstrReverse(&seqstr);
    }
    else
        ajStrAssignSubS(&seqstr, ajSeqGetSeqS(seq),
			begin-1,adj-1);

    ajStrFmtUpper(&seqstr);
    /*ajDebug("seqlen:%d len: %d offset: %d offend: %d begin: %d end: %d\n"
	   "'%S'\n",
	   seqlen , ajSeqGetLen(seq), ajSeqGetOffset(seq),
	   ajSeqGetOffend(seq), ajSeqGetBegin(seq), ajSeqGetEnd(seq),
	   seqstr);*/

    ajDebug("embPatternSeqSearch '%S' protein: %B reverse: %B\n",
	    pattern->pattern, pat->Protein, reverse);
    embPatFuzzSearchII(pattern,begin,seqname,seqstr,list,
                       ajPatternSeqGetMismatch(pat),&hits,&tidy);

    ajDebug ("embPatternSeqSearch: found %d hits\n",hits);

    if(!reverse)
	ajListReverse(list);

    for(i=0;i<hits;++i)
    {
        ajListPop(list,(void **)&m);

 	if (reverse)
	    sf = ajFeatNew(ftable, NULL, featMotifNuc,
                           adj - m->start - m->len + begin + 1,
                           adj - m->start + begin,
                           0.0, '-', 0);
	else
        {
	    if(ajSeqIsProt(seq) || ajFeattableIsProt(ftable))
                sf = ajFeatNewProt(ftable, NULL, featMotifProt,
                                   m->start,
                                   m->start + m->len - 1,
                                   0.0);
            else
                sf = ajFeatNew(ftable, NULL, featMotifNuc,
                               m->start,
                               m->start + m->len - 1,
                               0.0, '.', 0);
        }
        
	if(isreversed)
	    ajFeatReverse(sf, seqlen);

	/*
	ajUser("isrev: %B reverse: %B begin: %d adj: %d "
	       "start: %d len: %d seqlen: %d %d..%d '%c'\n",
	       isreversed, reverse, begin, adj, m->start, m->len, seqlen,
	       sf->Start, sf->End, sf->Strand);
	*/

	ajFeatSetScore(sf, (float) (m->len - m->mm));

        ajFmtPrintS(&tmp, "*pat %S: %S",
                    ajPatternSeqGetName(pat),
                    ajPatternSeqGetPattern(pat));
        ajFeatTagAdd(sf,NULL,tmp);

        if(m->mm)
        {
            ajFmtPrintS(&tmp, "*mismatch %d", m->mm);
            ajFeatTagAdd(sf, NULL, tmp);
        }

        embMatMatchDel(&m);
    }

    ajStrDel(&seqname);
    ajStrDel(&seqstr);
    ajStrDel(&tmp);
    ajListFree(&list);

    if (reverse)
        ajSeqDel(&revseq);

    return;
}